def check_seq_tax_labels(self, seq_name, orig_ranks, ranks, lws):
mislabel_lvl = -1
min_len = min(len(orig_ranks),len(ranks))
for rank_lvl in range(min_len):
if ranks[rank_lvl] != Taxonomy.EMPTY_RANK and ranks[rank_lvl] != orig_ranks[rank_lvl]:
mislabel_lvl = rank_lvl
break
if mislabel_lvl >= 0:
real_lvl = self.guess_rank_level(orig_ranks, mislabel_lvl)
mis_rec = {}
mis_rec['name'] = EpacConfig.strip_ref_prefix(seq_name)
mis_rec['orig_level'] = mislabel_lvl
mis_rec['real_level'] = real_lvl
mis_rec['level_name'] = self.rank_level_name(real_lvl)[1]
mis_rec['inv_level'] = -1 * real_lvl # just for sorting
mis_rec['orig_ranks'] = orig_ranks
mis_rec['ranks'] = ranks
mis_rec['lws'] = lws
mis_rec['conf'] = lws[mislabel_lvl]
self.mislabels.append(mis_rec)
return mis_rec
else:
return None
def check_seq_ids(self):
# check that seq IDs in taxonomy and alignment correspond
self.mis_ids = []
for sid in self.taxonomy.seq_ranks_map.iterkeys():
unprefixed_sid = EpacConfig.strip_ref_prefix(sid)
if not self.alignment.has_seq(unprefixed_sid):
self.mis_ids.append(unprefixed_sid)
if len(self.mis_ids) > 0 and self.verbose:
errmsg = "ERROR: Following %d sequence(s) are missing in your alignment file:\n%s\n\n" % (len(self.mis_ids), "\n".join(self.mis_ids))
errmsg += "Please make sure sequence IDs in taxonomic annotation file and in alignment are identical!\n"
self.cfg.exit_user_error(errmsg)
return self.mis_ids
def check_seq_ids(self):
# check that seq IDs in taxonomy and alignment correspond
self.mis_ids = []
for sid in self.taxonomy.seq_ranks_map.iterkeys():
unprefixed_sid = EpacConfig.strip_ref_prefix(sid)
if not self.alignment.has_seq(unprefixed_sid):
self.mis_ids.append(unprefixed_sid)
if len(self.mis_ids) > 0 and self.verbose:
errmsg = "ERROR: Following %d sequence(s) are missing in your alignment file:\n%s\n\n" % (
len(self.mis_ids), "\n".join(self.mis_ids))
errmsg += "Please make sure sequence IDs in taxonomic annotation file and in alignment are identical!\n"
self.cfg.exit_user_error(errmsg)
return self.mis_ids
def mis_rec_to_string(self, mis_rec):
lvl = mis_rec['orig_level']
uncorr_name = EpacConfig.strip_ref_prefix(self.refjson.get_uncorr_seqid(mis_rec['name']))
uncorr_orig_ranks = self.refjson.get_uncorr_ranks(mis_rec['orig_ranks'])
uncorr_ranks = self.refjson.get_uncorr_ranks(mis_rec['ranks'])
output = uncorr_name + "\t"
if lvl >= 0:
output += "%s\t%s\t%s\t%.3f\t" % (mis_rec['level_name'],
uncorr_orig_ranks[lvl], uncorr_ranks[lvl], mis_rec['lws'][lvl])
else:
output += "%s\t%s\t%s\t%.3f\t" % (mis_rec['level_name'],
"NA", "NA", mis_rec['lws'][0])
output += Taxonomy.lineage_str(uncorr_orig_ranks) + "\t"
output += Taxonomy.lineage_str(uncorr_ranks) + "\t"
output += ";".join(["%.3f" % conf for conf in mis_rec['lws']])
if 'rank_conf' in mis_rec:
output += "\t%.3f" % mis_rec['rank_conf']
return output
def mis_rec_to_string(self, mis_rec):
lvl = mis_rec['orig_level']
uncorr_name = EpacConfig.strip_ref_prefix(
self.refjson.get_uncorr_seqid(mis_rec['name']))
uncorr_orig_ranks = self.refjson.get_uncorr_ranks(
mis_rec['orig_ranks'])
uncorr_ranks = self.refjson.get_uncorr_ranks(mis_rec['ranks'])
output = uncorr_name + "\t"
if lvl >= 0:
output += "%s\t%s\t%s\t%.3f\t" % (
mis_rec['level_name'], uncorr_orig_ranks[lvl],
uncorr_ranks[lvl], mis_rec['lws'][lvl])
else:
output += "%s\t%s\t%s\t%.3f\t" % (mis_rec['level_name'], "NA",
"NA", mis_rec['lws'][0])
output += Taxonomy.lineage_str(uncorr_orig_ranks) + "\t"
output += Taxonomy.lineage_str(uncorr_ranks) + "\t"
output += ";".join(["%.3f" % conf for conf in mis_rec['lws']])
if 'rank_conf' in mis_rec:
output += "\t%.3f" % mis_rec['rank_conf']
return output
